Data processing device capable of performing problem diagnosis in a production system with plurality of robots and method

Abstract

A data processing device capable of performing problem diagnosis in a production system with a plurality of robots includes: a first time series obtaining part for obtaining historical event data used for determining some historical alarm indicator in time series and storing the historical event data as first time series data; a historic alarm indicator calculation part for calculating a series of historic alarm indicators using statistic characteristics of the first time series data; a threshold definition part for defining at least one threshold value based on a statistical distribution of the historical alarm indicators; a second time series obtaining part for obtaining operational event data during operation of the robots used for determining some operational alarm indicator in time series and storing the operational event data as second time series data; and an operational alarm indicator calculation part for calculating a series of operational alarm indicators.

What is claimed is: 
     
       1. A data processing device capable of performing problem diagnosis in a production system with a plurality of robots, comprising:
 a controller configured to:
 obtain historical event data used for determining some historical alarm indicator in time series and to store the historical event data as a first time series data; 
 calculate a series of historic alarm indicators using statistic characteristics of the first time series data; 
 define at least one threshold value based on a statistical distribution of the historical alarm indicators; 
 obtain operational event data during operation of the robots used for determining some operational alarm indicator in time series and to store the operational event data as a second time series data; 
 calculate a series of operational alarm indicators using statistic characteristics of the second time series data; 
 give alarm notifications to one of operational, maintenance, or troubleshooting personnel for alarm indicators above the at least one threshold level; and 
 highlight to one of the operational, maintenance, or troubleshooting personnel the events that mainly contribute to the operational alarm indicator by:
 highlighting a robot from the plurality of robots, associated with the operational alarm indicator that exceeded the at least one threshold; and 
 selecting additional information related to the highlighted robot that includes the current condition of the highlighted robot, and a list of events that caused the highlighted robot to exceed the threshold. 
 
 
 
     
     
       2. The data processing device according to  claim 1 , wherein the controller is further configured to:
 execute a machine learning algorithm in order to detect anomalies in the event data, wherein the event data comprises the historical event data and operational event data; and 
 identify events that probably cause a decision of the machine learning algorithm towards anomaly. 
 
     
     
       3. The data processing device according to  claim 1 , wherein the controller is further configured to include arbitrary input data comprising I/O or analog signals as event data. 
     
     
       4. A state-based or event based alarm system, comprising:
 the data processing device according to  claim 1 , which is configured, when state-based, to trigger an alarm while a score is above a threshold, and when event-based, to trigger an alarm when the alarm exceeds a threshold. 
 
     
     
       5. The data processing device according to  claim 1 , wherein the controller is configured to obtain historical event data in the first time series by collecting historic alarm indicators in a pre-defined window of time. 
     
     
       6. The data processing device according to  claim 1 , wherein the controller is configured to give alarm notifications by:
 calculating a difference between the obtained operational event data from the second time series data and at least one threshold value; and 
 generating a first kind of alarm based on the calculated difference. 
 
     
     
       7. The data processing device according to  claim 1 , wherein the respective statistic value for each respective historic alarm indicator in the series of historic alarm indicators is generated based on the statistic characteristics of the first time series data corresponding to the respective number of events in the historical event data in the respective observation period. 
     
     
       8. The data processing device according to  claim 1 , wherein the respective statistic value for each respective operational alarm indicator in the series of operational alarm indicators is generated based on the statistic characteristics of the second time series data corresponding to the respective number of events in the operational event data in the respective observation period. 
     
     
       9. The data processing device according to  claim 1 , wherein the respective historic alarm indicator of the series of historic alarm indicators corresponding to the respective observation period is calculated by: 
       
         
           
             
               ai 
               = 
               
                 
                   
                     
                       ❘ 
                       "\[LeftBracketingBar]" 
                     
                     
                       
                         x 
                         ′ 
                       
                       - 
                       
                         x 
                         ¯ 
                       
                     
                     
                       ❘ 
                       "\[RightBracketingBar]" 
                     
                   
                   - 
                   σ 
                 
                 
                   
                     ❘ 
                     "\[LeftBracketingBar]" 
                   
                   
                     
                       x 
                       ′ 
                     
                     - 
                     
                       x 
                       ¯ 
                     
                   
                   
                     ❘ 
                     "\[RightBracketingBar]" 
                   
                 
               
             
           
         
         wherein x′ is the respective number of events in the respective observation period,  x  is an average number of events, and σ is a standard deviation in number of events, 
         wherein the at least one threshold level comprises a first threshold level set to be satisfied when ai is above a first value, a second threshold level set to be satisfied when ai is above a second value, the second value being higher than the first value, and a third threshold level set to be satisfied when ai is above a third value, higher than the second value. 
       
     
     
       10. The data processing device according to  claim 1 , wherein each historic alarm indicator is a respective statistic value over a number of events in the historical event data in a respective observation period. 
     
     
       11. The data processing device according to  claim 10 , wherein each operational alarm indicator is a respective statistic value over a number of events in the operational event data in a respective observation period. 
     
     
       12. A method for performing unsupervised diagnosis in a production system with a plurality of robots, the method comprising steps of:
 S 1 : calculating historic alarm indicator values based on historical data, wherein each historic alarm indicator is a respective statistic value over a number of events in the historical event data in a respective observation period; 
 S 2 : defining and selecting a threshold value based on the distribution of the historic alarm indicator values; 
 S 3 : calculating an operational alarm indicator during live operation, using new event data, wherein each operational alarm indicator is a respective statistic value over a number of events in the operational event data in a respective observation period; 
 S 4 : giving alarm notifications for values above the threshold value, to operational, maintenance, or troubleshooting personnel; and 
 S 5 : highlighting to the operational, maintenance, or troubleshooting personnel the events that mainly contribute to the operational alarm indicator by:
 highlighting a robot from the plurality of robots, associated with the operational alarm indicator that exceeded the at least one threshold; and 
 selecting additional information related to the highlighted robot that includes the current condition of the highlighted robot, and a list of events that caused the highlighted robot to exceed the threshold. 
 
 
     
     
       13. The method according to  claim 12 , further comprising steps of:
 S 6 : running a machine learning algorithm to detect anomalies in the event data, wherein the event data comprises the historical event data and operational event data; and 
 S 7 : running an alarm indicator algorithm to identify the events that probably cause a decision of the machine learning algorithm towards anomaly. 
 
     
     
       14. The method according to  claim 12 , wherein the event data further comprises arbitrary input data comprising I/O or analog signals. 
     
     
       15. The method according to  claim 12 , wherein calculating the series of historic alarm indicator values is performed by collecting historic alarm indicators in a pre-defined window of time. 
     
     
       16. The method according to  claim 12 , wherein giving alarm notifications for values above the threshold value further comprises:
 calculating a difference between the operational event data from the second time series data and at least one threshold value; and 
 generating a first kind of alarm based on the calculated difference. 
 
     
     
       17. The method according to  claim 12 , wherein each historic alarm indicator is a respective statistic value over a number of events in the historical event data in a respective observation period. 
     
     
       18. The method according to  claim 17 , wherein each operational alarm indicator is a respective statistic value over a number of events in the operational event data in a respective observation period.